Forgot your password?
typodupeerror

Comment: Re:I know exactly what happened. (Score 1) 710

My two cents, freely speculating here until we get more facts:

This sounds like the founder's wife is a loose cannon with a her own little unofficial organization within the company. I have seen this before. This seems like the founder's wife was trying to recruit her into her network of spies.

Wouldn't it be more plausible that the founder (Asshole Entrepreneur or AE for short) and his wife (Colossal Bitch or CB) were teaming up to create the `spy network'? It would explain why AE let her come to the office when she had no bussiness being there. AE let CB do the dirty work so that he has plausible deniability.

This is where things are going sideways and neither she nor her partner see what is going on. By Horvath's partner talking to the founder's wife, they both made it onto her enemy list and became targets.

Indeed. Horvath had turned down the offer, and was now a liability because she knew too much. Partner talks to AE, now he's also a threat to both.

This is an indication that HR has been made aware of a situation and is investigating it. This was probably initiated by the founder's wife via the founder because of Horvath's partner.

AE and CB want to fire the couple and sic the HR department on them.

This is the investigation.

"Investigation" = "thinking of ways to get rid of the couple".

She said that having her personal relationship dragged into her work life and put on show for her coworkers didn’t sit well with her.

That is always a danger when one dates or is married to a coworker

AE and CB see this as a weakness they can exploit. The colleague making a pass at Horvath is a spy for CB and clumsily attempts to drive a wedge between Horvath and partner. Or maybe provoke a complaint to HR, which AE might then be able to twist into a reason for firing the couple.

The meeting did not go well.

>If she thought it would, she was a fool

She didn't, that's why she wanted HR there. She was a fool for thinking that HR would protect her against AE though. He's their paymaster, so they do what he says (probably a good cop/bad cop routine).

Never trust HR slime! They won't stick their neck out for you, and even if they did, it'd be chopped off. They don't have any real power.

Did she ask when she supposedly did that or ask for evidence?

Horvath cried during the episode,

What? Seriously? Instead of acting offended she acted like a child.

She probably didn't get a chance to talk about evidence or anything else, This was not a conversation, just a shouting / bullying session. Not everybody can deal with that kind of psychological pressure.

It is a bad idea to to date coworkers.

It seems they had been dating for a while, so why didn't AE bring it up much earlier? He was just looking for weak spots.

This is why I think the founder's wife said Horvath's partner delivered threat or something similar. That iis the most likely scenario. There is no proof, so he can't be fired for it, and the probably have no paper trail to fire him. But, I have little doubt they are building one.

I don't think threats were involved, because they'd have fired him on the spot. There is probably no paper trail for anything, because managers like this don't like to commit *anything* to paper. Why document your own diabolical schemes and incompetence and build a case against yourself?

Regardless of what Horvath thinks, this isn't about her being female. It is about power plays by the founder and his wife. Horvath and her partner were too blind to see what was going on. The founder's wife wanted Horvath on her "team" and things would have been OK if Horvath and her partner had just kept their mouths shut.

I agree that this is not about sexism. But things would never have been OK, even if she had agreed to spy. You just can't win if you're a pawn in a psycho's power game, best option is to update your resume and start applying for another job (from a home PC!).

the founder's wife engaged in a campaign to get rid of Horvath, the ungrateful person who had not just rejected being part of the founder's wife's circle but had her boyfriend tell the founder's wife to leave Horvath alone.

They went after them not for being ungrateful, but because they could compromise the spy network. Since the AE has now been sent on leave, it seems that the other founder didn't know about the spies. So probably the network was built to undermine his position.

is just Horvath being overly-sensitive probably because of the all the other stuff

She might not have been overly sensitive here, just (mis)calculating. The reason we're discussing this here on Slashdot is that she played the sexism card. If she had just complained about power games nobody would have noticed. This looks like a crude attempt at damaging the company's reputation. Very, very bad idea, who'd hire her now?

In the end everybody lost, but clearly AE and CB got what they deserved. It seems they love intrigue.. but they're not very good at it! They're unemployable IMHO. Still, some folks don't keep up with tech news and are impressed by titles like `co-founder of GitHub', so AE will probably be working at another tech company soon, and try the same stupid shit again.

Closing remarks:
I don't understand why some people are blaming the victim. She may have been naive, but she was in a difficult situation. I'm not sure what she could have done to avoid all this, except maybe avoid CB from the start. Another approach would have been to talk to the CEO about the spies, but it didn't occur to her because she was in the dark (we're all still speculating). Even then she might have been thrown out, we all know what happens to whistleblowers.

Comment: Re:NP-hard? (Score 1) 172

by DogPhilosopher (#45949407) Attached to: Regex Golf, xkcd, and Peter Norvig

.Giving a shitty algorithm to solve a problem, and then saying that it would also solve SET COVER, does not imply that any algorithm for solving the first problem can be used to solve SET COVER via a Karp reduction.

Btw, SET COVER is relevant for some learning problems. Constructing a fastest learner for finite identification of a finite class of finite languages (`preset learner') is equivalent to SET COVER, see page 6 of
http://staff.science.uva.nl/~dickdj/submission_COLT10_NGDdJ.pdf

Comment: Re:NP-hard? (Score 1) 172

by DogPhilosopher (#45944441) Attached to: Regex Golf, xkcd, and Peter Norvig

Sure, it's NP-hard, but it's also nonrecursive via Rice's theorem.

That depends a bit on how the problem is formulated. When stated in full generality, ie "given an arbitrary UTM, decide whether it's a regex finder", it's obviously not decidable.

However, the full expressive power of Turing machines may not be needed. For some classes of grammars/automata, the question whether something is a member of the class can be decided by an automaton, this property is known as automaticity and has origins in the field of descriptive complexity. See for example
http://semukhin.name/autofeed2.pdf

Also relevant, work on the enumeration of regexps, eg
http://arxiv.org/abs/1204.4982

Comment: Re:NP-hard? (Score 1) 172

by DogPhilosopher (#45937897) Attached to: Regex Golf, xkcd, and Peter Norvig

Giving a shitty algorithm to solve a problem, and then saying that it would also solve SET COVER, does not imply that any algorithm for solving the first problem can be used to solve SET COVER via a Karp reduction.

Indeed. See my posts above for a reference to a Karp reduction from VERTEX COLORING (for the automaton version). Previous hardness and non-approximability results for this and related problems were found by Gold (early 70s!), Angluin and Abe.

Myabe also worth mentioning that not all NP-hard problems are created equal. The natural parametrization of SET COVER is W[2]-hard, but there are no decent re-parametrizations known for coloring that give it a place in the W-hierarchy. I know only of one trivial one.

Coloring is still NP-hard when graphs are restricted to be 3-colorable, planar and having maximum degree 4. it seems to be a very hard problem in certain ways.

Comment: Re:This problem has been studied for decades (Score 1) 172

by DogPhilosopher (#45937787) Attached to: Regex Golf, xkcd, and Peter Norvig

I guess you're saying that it's still a legitimate subject, and that progress is made by building on previous results. Nobody is disputing that.

But the TFA doesn't mention previous results or even the existence of the field, I have the impression that Norvig is not aware of it. So this is not contributing to the body of knowledge re automata induction, this is recreative CS. Nothing wrong with that per se, but there's also nothing wrong with providing some scientific background.

Btw in the blog post, the approximative approach is motivated by NP-hardness of the exact problem. Given the link (parameter-preserving reduction) with graph coloring already mentioned, the problem can't be approximated in polynomial time with arbitrary error, unless P=NP. This theoretical result is backed up by practical experiments with approximating coloring algorithms, which often find solutions 100% off from the optimal nr of colors. So good luck with that.

Also, informed regular inference is already NP-hard when restricted to DFAs with just 2 states:
http://link.springer.com/chapter/10.1007%2F978-3-642-37064-9_25

Perhaps surprisingly, the problem becomes tractable when the data is `complete', in the sense that it is consistent with just one single automaton (google RPNI).

Comment: This problem has been studied for decades (Score 5, Informative) 172

by DogPhilosopher (#45933955) Attached to: Regex Golf, xkcd, and Peter Norvig

There's a field called Grammar Induction, and the problem of learning regular languages, aka regular inference, can be considered a subfield. People have been working on this since the '50s. Applications include learning DTDs for XML/wrapper induction, and all kinds of problems in bioinformatics and natural language processing.

There's a strong link with the graph coloring problem, see
http://www.cs.ru.nl/~sicco/papers/alt12.pdf

In this field, the focus is generally on learning FSAs, but these can easily be transformed into regexps. There's work on learning regexps directly, see
http://www.informatik.uni-trier.de/~fernau/papers/Fer05c.pdf

Enjoy.

Businesses

High-Frequency Traders Use 50-Year-Old Wireless Tech 395

Posted by Unknown Lamer
from the get-with-the-times dept.
jfruh writes "In the world of high-frequency stock trading, every millisecond is money. That's why many firms are getting information and sending big orders not through modern fiber-optic networks, but using line-of-site microwave repeaters, a technology that's over 50 years old. Because electromagnetic radiation passes more quickly through air than glass, and takes a more direct route, the older technology is seeing something of a renaissance."

Memory fault -- brain fried

Working...